Eleven healthy volunteers received a single intravenous dose of diazepam (0.15 mg/kg), midazolam (0.1 mg/kg), and placebo by 1-minute infusion in a double-blind, three-way crossover study. Plasma concentrations were measured during 24 hours after dosage, and the electroencephalographic (EEG) power spectrum was simultaneously computed by fast-Fourier transform to determine the percentage of total EEG amplitude occurring in the 13 to 30 Hz range. Both diazepam and midazolam had large volumes of distribution (1.2 and 2.3 L/kg, respectively), but diazepam's half-life was considerably longer (33 versus 2.8 hours) and its metabolic clearance lower (0.5 versus 11.0 ml/min kg) than those of midazolam. EEG changes were maximal at the end of the diazepam infusion and 5 to 10 minutes after midazolam infusion. Percent 13 to 30 Hz activity remained significantly above baseline until 5 hours for diazepam but only until 2 hours for midazolam. For both drugs, EEG effects were indistinguishable from baseline by 6 to 8 hours, suggesting that distribution contributes importantly to terminating pharmacodynamic action. The relationship of EEG change to plasma drug concentration indicated an apparent EC50 value of 269 ng/ml for diazepam as opposed to 35 ng/ml for midazolam. However, Emax values were similar for both drugs (+19.4% and +21.3%, respectively).